Code

data: Added support for a "regex" data-type.
[sysdb.git] / src / core / data.c
1 /*
2  * SysDB - src/core/data.c
3  * Copyright (C) 2014 Sebastian 'tokkee' Harl <sh@tokkee.org>
4  * All rights reserved.
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
16  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
17  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
18  * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR
19  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
20  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
21  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
22  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
23  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
24  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
25  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26  */
28 #if HAVE_CONFIG_H
29 #       include "config.h"
30 #endif /* HAVE_CONFIG_H */
32 #include "sysdb.h"
34 #include "core/data.h"
35 #include "utils/error.h"
37 #include <errno.h>
39 #include <inttypes.h>
41 #include <stdio.h>
42 #include <stdlib.h>
43 #include <string.h>
45 #include <math.h>
47 /*
48  * private helper functions
49  */
51 /* Calculate the linear function 'd1 + n * d2'. */
52 static int
53 data_lin(const sdb_data_t *d1, int n, const sdb_data_t *d2, sdb_data_t *res)
54 {
55         if (d1->type != d2->type)
56                 return -1;
58         if (d1->type == SDB_TYPE_INTEGER)
59                 res->data.integer = d1->data.integer + (int64_t)n * d2->data.integer;
60         else if (d1->type == SDB_TYPE_DECIMAL)
61                 res->data.decimal = d1->data.decimal + (double)n * d2->data.decimal;
62         else if (d1->type ==  SDB_TYPE_DATETIME)
63                 res->data.datetime = d1->data.datetime + (sdb_time_t)n * d2->data.datetime;
64         else
65                 return -1;
66         res->type = d1->type;
67         return 0;
68 } /* data_lin */
70 /* Multiply d1 with d2. */
71 static int
72 data_mul(const sdb_data_t *d1, const sdb_data_t *d2, sdb_data_t *res)
73 {
74         if (d1->type == SDB_TYPE_INTEGER) {
75                 if (d2->type == SDB_TYPE_INTEGER)
76                         res->data.integer = d1->data.integer * d2->data.integer;
77                 else if (d2->type == SDB_TYPE_DATETIME) {
78                         res->data.datetime = (sdb_time_t)d1->data.integer
79                                 * d2->data.datetime;
80                         res->type = SDB_TYPE_DATETIME;
81                         return 0;
82                 }
83                 else
84                         return -1;
85         }
86         else if (d1->type == SDB_TYPE_DECIMAL) {
87                 if (d2->type == SDB_TYPE_DECIMAL)
88                         res->data.decimal = d1->data.decimal * d2->data.decimal;
89                 else if (d2->type == SDB_TYPE_DATETIME) {
90                         res->data.datetime = (sdb_time_t)(d1->data.decimal
91                                         * (double)d2->data.datetime);
92                         res->type = SDB_TYPE_DATETIME;
93                         return 0;
94                 }
95                 else
96                         return -1;
97         }
98         else if (d1->type == SDB_TYPE_DATETIME) {
99                 if (d2->type == SDB_TYPE_DATETIME)
100                         res->data.datetime = d1->data.datetime
101                                 * d2->data.datetime;
102                 else if (d2->type == SDB_TYPE_INTEGER)
103                         res->data.datetime = d1->data.datetime
104                                 * (sdb_time_t)d2->data.integer;
105                 else if (d2->type == SDB_TYPE_DECIMAL)
106                         res->data.datetime = (sdb_time_t)((double)d1->data.datetime
107                                         * d2->data.decimal);
108                 else
109                         return -1;
110         }
111         else
112                 return -1;
114         res->type = d1->type;
115         return 0;
116 } /* data_mul */
118 /* Device d1 by d2 and return the result and the remainder. */
119 static int
120 data_div(const sdb_data_t *d1, const sdb_data_t *d2,
121                 sdb_data_t *res, sdb_data_t *rem)
123         if (d1->type == SDB_TYPE_INTEGER) {
124                 if (d2->type != SDB_TYPE_INTEGER)
125                         return -1;
126                 if (res)
127                         res->data.integer = d1->data.integer / d2->data.integer;
128                 if (rem)
129                         rem->data.integer = d1->data.integer % d2->data.integer;
130         }
131         else if (d1->type == SDB_TYPE_DECIMAL) {
132                 if (d2->type != SDB_TYPE_DECIMAL)
133                         return -1;
134                 if (res)
135                         res->data.decimal = d1->data.decimal / d2->data.decimal;
136                 if (rem)
137                         rem->data.decimal = fmod(d1->data.decimal, d2->data.decimal);
138         }
139         else if (d1->type == SDB_TYPE_DATETIME) {
140                 if (d2->type == SDB_TYPE_DECIMAL) {
141                         if (res)
142                                 res->data.datetime = (sdb_time_t)((double)d1->data.datetime
143                                                 / d2->data.decimal);
144                         if (rem) {
145                                 double tmp = fmod((double)d1->data.datetime, d2->data.decimal);
146                                 rem->data.datetime = (sdb_time_t)tmp;
147                         }
148                 }
149                 else {
150                         sdb_time_t a, b;
151                         if (d2->type == SDB_TYPE_DATETIME) {
152                                 a = d1->data.datetime;
153                                 b = d2->data.datetime;
154                         }
155                         else if (d2->type == SDB_TYPE_INTEGER) {
156                                 a = d1->data.datetime;
157                                 b = (sdb_time_t)d2->data.integer;
158                         }
159                         else
160                                 return -1;
161                         if (res)
162                                 res->data.datetime = a / b;
163                         if (rem)
164                                 rem->data.datetime = a % b;
165                 }
166         }
167         else
168                 return -1;
170         if (res)
171                 res->type = d1->type;
172         if (rem)
173                 rem->type = d1->type;
174         return 0;
175 } /* data_div */
177 /* Concatenate d1 and d2. */
178 static int
179 data_concat(const sdb_data_t *d1, const sdb_data_t *d2, sdb_data_t *res)
181         unsigned char *new;
182         unsigned char *s1, *s2;
183         size_t len1, len2;
185         if (d1->type != d2->type)
186                 return -1;
188         if (d1->type == SDB_TYPE_STRING) {
189                 s1 = (unsigned char *)d1->data.string;
190                 s2 = (unsigned char *)d2->data.string;
191                 len1 = s1 ? strlen((char *)s1) : 0;
192                 len2 = s2 ? strlen((char *)s2) : 0;
193         }
194         else if (d1->type == SDB_TYPE_BINARY) {
195                 s1 = d1->data.binary.datum;
196                 s2 = d2->data.binary.datum;
197                 len1 = d1->data.binary.length;
198                 len2 = d2->data.binary.length;
199         }
200         else
201                 return -1;
203         if (s1 || s2) {
204                 new = malloc(len1 + len2 + 1);
205                 if (! new)
206                         return -1;
207         }
208         else
209                 new = NULL;
211         if (len1)
212                 memcpy(new, s1, len1);
213         if (len2)
214                 memcpy(new + len1, s2, len2);
215         if (new)
216                 new[len1 + len2] = '\0';
218         res->type = d1->type;
219         if (res->type == SDB_TYPE_STRING) {
220                 res->data.string = (char *)new;
221         }
222         else {
223                 res->data.binary.datum = new;
224                 res->data.binary.length = len1 + len2;
225         }
226         return 0;
227 } /* data_concat */
229 /*
230  * public API
231  */
233 int
234 sdb_data_copy(sdb_data_t *dst, const sdb_data_t *src)
236         sdb_data_t tmp;
238         if ((! dst) || (! src))
239                 return -1;
241         tmp = *src;
242         switch (src->type) {
243                 case SDB_TYPE_STRING:
244                         if (src->data.string) {
245                                 tmp.data.string = strdup(src->data.string);
246                                 if (! tmp.data.string)
247                                         return -1;
248                         }
249                         break;
250                 case SDB_TYPE_BINARY:
251                         if (src->data.binary.datum) {
252                                 tmp.data.binary.datum = malloc(src->data.binary.length);
253                                 if (! tmp.data.binary.datum)
254                                         return -1;
255                                 memcpy(tmp.data.binary.datum, src->data.binary.datum,
256                                                 src->data.binary.length);
257                         }
258                         break;
259                 case SDB_TYPE_REGEX:
260                         if (src->data.re.raw) {
261                                 tmp.data.re.raw = strdup(src->data.re.raw);
262                                 if (! tmp.data.re.raw)
263                                         return -1;
264                                 /* we need to recompile because the regex might point to
265                                  * dynamically allocated memory */
266                                 if (regcomp(&tmp.data.re.regex, tmp.data.re.raw,
267                                                         REG_EXTENDED | REG_ICASE | REG_NOSUB)) {
268                                         free(tmp.data.re.raw);
269                                         return -1;
270                                 }
271                         }
272                         else
273                                 memset(&tmp.data.re.regex, 0, sizeof(tmp.data.re.regex));
274                         break;
275         }
277         sdb_data_free_datum(dst);
278         *dst = tmp;
279         return 0;
280 } /* sdb_data_copy */
282 void
283 sdb_data_free_datum(sdb_data_t *datum)
285         if (! datum)
286                 return;
288         switch (datum->type) {
289                 case SDB_TYPE_STRING:
290                         if (datum->data.string)
291                                 free(datum->data.string);
292                         datum->data.string = NULL;
293                         break;
294                 case SDB_TYPE_BINARY:
295                         if (datum->data.binary.datum)
296                                 free(datum->data.binary.datum);
297                         datum->data.binary.datum = NULL;
298                         datum->data.binary.length = 0;
299                         break;
300                 case SDB_TYPE_REGEX:
301                         if (datum->data.re.raw) {
302                                 free(datum->data.re.raw);
303                                 regfree(&datum->data.re.regex);
304                         }
305                         datum->data.re.raw = NULL;
306                         memset(&datum->data.re.regex, 0, sizeof(datum->data.re.regex));
307                         break;
308         }
309 } /* sdb_data_free_datum */
311 int
312 sdb_data_cmp(const sdb_data_t *d1, const sdb_data_t *d2)
314 #define CMP_NULL(a, b) \
315         do { \
316                 if (!(a) && !(b)) return 0; \
317                 if (!(a)) return -1; \
318                 if (!(b)) return 1; \
319         } while (0)
321         CMP_NULL(d1, d2);
323         if (d1->type != d2->type)
324                 return SDB_CMP(d1->type, d2->type);
326         switch (d1->type) {
327                 case SDB_TYPE_INTEGER:
328                         return SDB_CMP(d1->data.integer, d2->data.integer);
329                 case SDB_TYPE_DECIMAL:
330                         return SDB_CMP(d1->data.decimal, d2->data.decimal);
331                 case SDB_TYPE_STRING:
332                         CMP_NULL(d1->data.string, d2->data.string);
333                         return strcasecmp(d1->data.string, d2->data.string);
334                 case SDB_TYPE_DATETIME:
335                         return SDB_CMP(d1->data.datetime, d2->data.datetime);
336                 case SDB_TYPE_BINARY:
337                 {
338                         int diff;
340                         CMP_NULL(d1->data.binary.datum, d2->data.binary.datum);
342                         /* on a common prefix, the shorter datum sorts less */
343                         if (d1->data.binary.length < d2->data.binary.length) {
344                                 diff = memcmp(d1->data.binary.datum, d2->data.binary.datum,
345                                                 d1->data.binary.length);
346                                 diff = diff ? diff : -1;
347                         }
348                         else if (d1->data.binary.length > d2->data.binary.length) {
349                                 diff = memcmp(d1->data.binary.datum, d2->data.binary.datum,
350                                                 d2->data.binary.length);
351                                 diff = diff ? diff : 1;
352                         }
353                         else
354                                 diff = memcmp(d1->data.binary.datum, d2->data.binary.datum,
355                                                 d1->data.binary.length);
357                         return diff;
358                 }
359                 case SDB_TYPE_REGEX:
360                         CMP_NULL(d1->data.re.raw, d2->data.re.raw);
361                         return strcmp(d1->data.re.raw, d2->data.re.raw);
362         }
363         return -1;
364 } /* sdb_data_cmp */
366 int
367 sdb_data_strcmp(const sdb_data_t *d1, const sdb_data_t *d2)
369         char d1_str[sdb_data_strlen(d1) + 1];
370         char d2_str[sdb_data_strlen(d2) + 1];
372         if (sdb_data_isnull(d1))
373                 d1 = NULL;
374         if (sdb_data_isnull(d2))
375                 d2 = NULL;
377         CMP_NULL(d1, d2);
379         if (sdb_data_format(d1, d1_str, sizeof(d1_str), SDB_UNQUOTED) < 0)
380                 return SDB_CMP(sizeof(d1_str), sizeof(d2_str));
381         if (sdb_data_format(d2, d2_str, sizeof(d2_str), SDB_UNQUOTED) < 0)
382                 return SDB_CMP(sizeof(d1_str), sizeof(d2_str));
384         return strcasecmp(d1_str, d2_str);
385 #undef CMP_NULL
386 } /* sdb_data_strcmp */
388 _Bool
389 sdb_data_isnull(const sdb_data_t *datum)
391         if (! datum)
392                 return 1;
393         if ((datum->type == SDB_TYPE_STRING) && (! datum->data.string))
394                 return 1;
395         if ((datum->type == SDB_TYPE_BINARY) && (! datum->data.binary.datum))
396                 return 1;
397         if ((datum->type == SDB_TYPE_REGEX) && (! datum->data.re.raw))
398                 return 1;
399         return 0;
400 } /* sdb_data_isnull */
402 int
403 sdb_data_expr_eval(int op, const sdb_data_t *d1, const sdb_data_t *d2,
404                 sdb_data_t *res)
406         if ((! d1) || (! d2) || (! res))
407                 return -1;
408         switch (op) {
409                 case SDB_DATA_CONCAT:
410                         return data_concat(d1, d2, res);
411                 case SDB_DATA_ADD:
412                         return data_lin(d1, 1, d2, res);
413                 case SDB_DATA_SUB:
414                         return data_lin(d1, -1, d2, res);
415                 case SDB_DATA_MUL:
416                         return data_mul(d1, d2, res);
417                 case SDB_DATA_DIV:
418                         return data_div(d1, d2, res, NULL);
419                 case SDB_DATA_MOD:
420                         return data_div(d1, d2, NULL, res);
421         }
422         return -1;
423 } /* sdb_data_expr_eval */
425 size_t
426 sdb_data_strlen(const sdb_data_t *datum)
428         if (! datum)
429                 return 0;
431         switch (datum->type) {
432                 case SDB_TYPE_INTEGER:
433                         /* log(64) */
434                         return 20;
435                 case SDB_TYPE_DECIMAL:
436                         /* XXX: -d.dddddde+dd or -ddddd.dddddd */
437                         return 42;
438                 case SDB_TYPE_STRING:
439                         if (! datum->data.string)
440                                 return 8; /* "<NULL>" */
441                         /* in the worst case, each character needs to be escaped */
442                         return 2 * strlen(datum->data.string) + 2;
443                 case SDB_TYPE_DATETIME:
444                         /* "YYYY-MM-DD HH:MM:SS +zzzz" */
445                         return 27;
446                 case SDB_TYPE_BINARY:
447                         if (! datum->data.binary.datum)
448                                 return 8; /* "<NULL>" */
449                         /* "\xNN" */
450                         return 4 * datum->data.binary.length + 2;
451                 case SDB_TYPE_REGEX:
452                         if (! datum->data.re.raw)
453                                 return 8; /* "<NULL>" */
454                         /* "/.../" */
455                         return strlen(datum->data.re.raw) + 4;
456         }
457         return 0;
458 } /* sdb_data_strlen */
460 int
461 sdb_data_format(const sdb_data_t *datum, char *buf, size_t buflen, int quoted)
463         char tmp[sdb_data_strlen(datum) + 1];
464         char *data = NULL;
465         int ret = -1;
467         size_t i, pos;
469         if ((! datum) || (! buf))
470                 return -1;
472         switch (datum->type) {
473                 case SDB_TYPE_INTEGER:
474                         ret = snprintf(buf, buflen, "%"PRIi64, datum->data.integer);
475                         break;
476                 case SDB_TYPE_DECIMAL:
477                         ret = snprintf(buf, buflen, "%g", datum->data.decimal);
478                         break;
479                 case SDB_TYPE_STRING:
480                         if (! datum->data.string)
481                                 data = "<NULL>";
482                         else {
483                                 pos = 0;
484                                 for (i = 0; i < strlen(datum->data.string); ++i) {
485                                         char byte = datum->data.string[i];
487                                         if ((byte == '\\') || (byte == '"')) {
488                                                 tmp[pos] = '\\';
489                                                 ++pos;
490                                         }
491                                         tmp[pos] = byte;
492                                         ++pos;
493                                 }
494                                 tmp[pos] = '\0';
495                                 data = tmp;
496                         }
497                         break;
498                 case SDB_TYPE_DATETIME:
499                         if (! sdb_strftime(tmp, sizeof(tmp), "%F %T %z",
500                                                 datum->data.datetime))
501                                 return -1;
502                         tmp[sizeof(tmp) - 1] = '\0';
503                         data = tmp;
504                         break;
505                 case SDB_TYPE_BINARY:
506                         pos = 0;
507                         for (i = 0; i < datum->data.binary.length; ++i) {
508                                 int byte = (int)datum->data.binary.datum[i];
509                                 char hex[] = {'0', '1', '2', '3', '4', '5', '6', '7',
510                                         '8', '9', 'a', 'b', 'c', 'd', 'e', 'f'};
512                                 tmp[pos] = '\\';
513                                 tmp[pos + 1] = 'x';
514                                 pos += 2;
516                                 if (byte > 0xf) {
517                                         tmp[pos] = hex[byte >> 4];
518                                         ++pos;
519                                 }
520                                 tmp[pos] = hex[byte & 0xf];
521                                 ++pos;
522                         }
523                         if (datum->data.binary.datum) {
524                                 tmp[pos] = '\0';
525                                 data = tmp;
526                         }
527                         else
528                                 data = "<NULL>";
529                         break;
530                 case SDB_TYPE_REGEX:
531                         if (! datum->data.re.raw)
532                                 data = "<NULL>";
533                         else {
534                                 snprintf(tmp, sizeof(tmp), "/%s/", datum->data.re.raw);
535                                 data = tmp;
536                         }
537                         break;
538         }
540         if (data) {
541                 if (quoted == SDB_UNQUOTED)
542                         ret = snprintf(buf, buflen, "%s", data);
543                 else if (quoted == SDB_SINGLE_QUOTED)
544                         ret = snprintf(buf, buflen, "'%s'", data);
545                 else
546                         ret = snprintf(buf, buflen, "\"%s\"", data);
547         }
548         buf[buflen - 1] = '\0';
549         return ret;
550 } /* sdb_data_format */
552 int
553 sdb_data_parse(char *str, int type, sdb_data_t *data)
555         sdb_data_t tmp;
557         char *endptr = NULL;
559         errno = 0;
560         switch (type) {
561                 case SDB_TYPE_INTEGER:
562                         tmp.data.integer = strtoll(str, &endptr, 0);
563                         break;
564                 case SDB_TYPE_DECIMAL:
565                         tmp.data.decimal = strtod(str, &endptr);
566                         break;
567                 case SDB_TYPE_STRING:
568                         tmp.data.string = str;
569                         break;
570                 case SDB_TYPE_DATETIME:
571                         {
572                                 double datetime = strtod(str, &endptr);
573                                 tmp.data.datetime = DOUBLE_TO_SDB_TIME(datetime);
574                         }
575                         break;
576                 case SDB_TYPE_BINARY:
577                         /* we don't support any binary information containing 0-bytes */
578                         tmp.data.binary.length = strlen(str);
579                         tmp.data.binary.datum = (unsigned char *)str;
580                         break;
581                 case SDB_TYPE_REGEX:
582                         tmp.data.re.raw = strdup(str);
583                         if (! tmp.data.re.raw)
584                                 return -1;
585                         if (regcomp(&tmp.data.re.regex, str,
586                                                 REG_EXTENDED | REG_ICASE | REG_NOSUB)) {
587                                 free(tmp.data.re.raw);
588                                 sdb_log(SDB_LOG_ERR, "core: Failed to compile regular "
589                                                 "expression '%s'", str);
590                                 return -1;
591                         }
592                         if (! data) {
593                                 tmp.type = SDB_TYPE_REGEX;
594                                 sdb_data_free_datum(&tmp);
595                         }
596                         break;
597                 default:
598                         errno = EINVAL;
599                         return -1;
600         }
602         if ((type == SDB_TYPE_INTEGER) || (type == SDB_TYPE_DECIMAL)
603                         || (type == SDB_TYPE_DATETIME)) {
604                 if (errno || (str == endptr)) {
605                         char errbuf[1024];
606                         sdb_log(SDB_LOG_ERR, "core: Failed to parse string "
607                                         "'%s' as numeric value (type %i): %s", str, type,
608                                         sdb_strerror(errno, errbuf, sizeof(errbuf)));
609                         return -1;
610                 }
611                 else if (endptr && (*endptr != '\0'))
612                         sdb_log(SDB_LOG_WARNING, "core: Ignoring garbage after "
613                                         "number while parsing numeric value (type %i): %s.",
614                                         type, endptr);
615         }
617         if (data) {
618                 *data = tmp;
619                 data->type = type;
620         }
621         return 0;
622 } /* sdb_data_parse */
624 /* vim: set tw=78 sw=4 ts=4 noexpandtab : */